The Psychology of Arithmetic

Thorndike argues that arithmetic is not a single mental faculty or a general power of reasoning that can be trained and then applied to any numerical problem. Instead, it is a collection of specific stimulus-response bonds: particular connections between number combinations, operations, and responses that must be taught directly. The book rejects the doctrine of formal discipline, which held that studying arithmetic strengthens general reasoning ability. Thorndike demonstrates that transfer between arithmetic topics is limited and that instruction must target the exact bonds students need to acquire. Arithmetic ability, on this account, is the sum of many particular connections, not the exercise of a unified cognitive capacity.

Thorndike analyzes arithmetic into several categories of bonds. There are bonds between number combinations and their sums, differences, products, and quotients—the basic facts that must become automatic. There are bonds between verbal statements of problems and the operations they require, such as recognizing that 'in all' calls for addition. There are bonds between arithmetic operations and their written symbols and procedures. There are bonds involving the sequence of steps in multi-step computations. And there are bonds connecting arithmetic knowledge to practical situations. Each of these must be learned separately; mastery of one does not guarantee mastery of another. This decomposition was intended to make instruction more precise and efficient.

The doctrine of formal discipline held that studying subjects like arithmetic and Latin strengthened general mental faculties such as reasoning, memory, and attention, which would then transfer to any domain. Thorndike's experiments on transfer found that learning transfers only insofar as the two tasks share identical elements. In arithmetic, this means that practicing addition does not automatically improve subtraction, and that learning to compute does not automatically improve the ability to solve word problems. These findings undermined the justification for teaching arithmetic primarily as mental discipline. Thorndike argued instead that arithmetic should be taught for its specific practical and intellectual uses, with instruction organized around the particular bonds students will need.

Thorndike offers several recommendations based on his bond theory. First, instruction should identify the specific bonds to be learned and teach them directly, rather than relying on incidental learning or general practice. Second, the bonds should be taught in the order of their usefulness and difficulty, with the most frequently needed and easiest connections established first. Third, practice should be distributed and should include immediate feedback, since the law of effect requires that correct responses be followed by satisfying consequences. Fourth, instruction should provide varied contexts so that bonds generalize appropriately. Fifth, teachers should measure student progress in terms of specific bonds acquired, not general impressions of ability. These recommendations were influential in the development of structured arithmetic curricula.

The book was widely read and became a foundational text for the scientific movement in arithmetic education. It was praised for its empirical rigor and for providing a concrete, testable alternative to vague appeals to mental discipline. Critics, however, questioned whether the decomposition of arithmetic into thousands of specific bonds was practical for classroom instruction and whether it neglected the conceptual understanding that underlies mathematical thinking. Despite these criticisms, the book had substantial influence on curriculum design, textbook writing, and the development of standardized arithmetic tests. Its emphasis on measuring specific learning outcomes anticipated the behavioral objectives movement that would later dominate educational planning, and its critique of formal discipline contributed to the decline of classical curricula in American schools.

The Psychology of Arithmetic is one of a series of Thorndike's applied works that translated his connectionist theory into specific educational domains. It applies the laws of learning—readiness, exercise, and effect—and the principle of identical elements to the concrete problem of teaching arithmetic. It is closely related to his three-volume Educational Psychology (1913–1914) and to his other subject-matter psychologies, including The Psychology of Algebra (1923) and The Teacher's Word Book (1921). Together these works represent the most systematic attempt to build an educational science on behaviorist principles before Skinner's programmed instruction. The book thus stands as a key document in the history of both learning theory and curriculum reform.